Skills Management in Unique AI Conduct

Overview

Skills are reusable capability packages that extend the Conduct agent's behavior for specific workflows.

Skills reach a session from two independent places:

• Knowledge Base skills are authored by your organization and stored as folders in the Knowledge Base. When a Conduct space has the Skills tool enabled and pointed at a Knowledge Base folder, the agent loads every skill in that folder at the start of every turn. If no folder is selected, the space runs on built-in skills only.

• Built-in skills ship with the product. These are not stored in the Knowledge Base. A few appear only when the matching feature is enabled on the space.

This article covers how to organize, manage, and create skills for your organization.

Who is it for?

Administrators and team leads responsible for configuring Conduct spaces and managing the skills library for their organization.

If you're unable to access the features discussed on this page, it's possible that your company does not have access or hasn't upgraded to the latest version yet. Please reach out to your internal support team or Unique's customer success team for assistance.

Permissions and Ownership

Managing skills involves two separate permissions:

• Manage access on the space — to enable the Skills tool and choose which Knowledge Base folder it loads. Held by the space owner, space managers, and company-wide space administrators.

• Write access on a Knowledge Base folder — to add or edit the skills inside it. This is what skill authors need, and it is granted per folder.

The two are independent: managing a space does not let you edit the skills it loads, and having write access to a skills folder does not let you point a space at it. Users need neither to save a personal skill from a session — that only requires skill creation to be enabled on the space, and the result lands in the user's own private layer.

We recommend assigning ownership per folder — company-wide folders to a small group of skills admins, team folders to team leads, use-case folders to their owners. These are organizational roles, implemented purely by granting Knowledge Base write access on the matching folders.

Skill Anatomy

Every skill is a folder containing at minimum a SKILL.md file:

my-skill/
├── SKILL.md           ← Required — manifest with instructions
├── scripts/           ← Optional — executable code for deterministic tasks
├── references/        ← Optional — docs loaded into context as needed
├── assets/            ← Optional — templates, icons, fonts
└── examples/          ← Optional — input/output examples

Any folder anywhere under the space's skills folder that contains a SKILL.md is treated as a skill, and its entire subtree is loaded with it. Grouping folders are therefore free to nest as deeply as you like.

SKILL.md Format

---
name: "financial-document-parser"
description: "Extracts key data points, tables, and metadata from financial documents (10-K, DDQ, side letters). Use when the user uploads a financial PDF or asks to parse/extract from a regulatory document."
---

[Detailed instructions for the agent — what to extract, how to format output,
edge cases to handle, quality checks to run]

The name and description in the front matter are critical — the description determines when the agent selects this skill. Make it specific about trigger conditions.

For a comprehensive guide on skill authoring — including best practices, structure patterns, and examples — visit agentskills.io.

Organizing Skills

Skills that your organization creates must be stored in the Knowledge Base. Conduct spaces load skills from a Knowledge Base folder (called skills-conductby default). The folders directly inside it are more than organization: four reserved name prefixes define precedence layers, which determine what happens when two skills share the same name (see the Skill Precedence subsection below). Use them to mirror how permissions and ownership flow in regulated environments. Below is an example of a folder structure:

skills-conduct/                      ← the default folder for Conduct spaces
├── company-core/                    [read: all]   [write: skills admins]
│   ├── house-style-memo/
│   ├── bloomberg-ticker-resolver/
│   └── compliance-disclaimer-injector/
│
├── space-Research Assistant/        [read: all]   [write: space owner]
│   └── (skills that apply only to the space with this exact name)
│
├── team-long-short-equity/          [read: team]  [write: team leads]
├── team-credit/
├── team-macro/
│
├── personal-<userId>/               [read+write: that user only]
│   └── (created automatically when a user saves a skill)
│
└── use-cases/                       ← a plain folder: organization only, no precedence
    ├── ddq-rfp/
    ├── 10-K-analysis/
    └── side-letter-extraction/

Skill Precedence

Skills are applied in the following order, from base to top. When a skill name appears in more than one layer, the highest layer wins outright — its folder replaces the lower layer's skill wholesale, including scripts/, references/, and assets/. Nothing is merged.

1. Built-in skills — shipped with the product (see below)

2. Plain folders — any folder without a reserved prefix

3. company-*

4. space-<space name> — only when the suffix matches the current space's name exactly

5. team-*

6. personal-<userId> — only the current user's own folder

Two things follow from this that are worth knowing before you name folders:

• personal-<userId> is a genuine privacy boundary. Other users' personal-* folders are never loaded, even when Knowledge Base permissions would allow reading them. This only works with the user ID — a folder named after a username (personal-j.smith) gets no boundary and is loaded for everyone in scope.

• A space-* folder whose suffix does not match the active space is treated as a plain folder rather than skipped. Its skills are still loaded for everyone with read access — they just carry no precedence. Check the space name spelling if you rely on space-level overrides.

Promoting a Skill to a Wider Audience

A skill that proves useful for one person is often worth sharing with a team or the whole company. To To promote a skill, copy the skill folder into the wider layer (team-* or company-*) and grant access accordingly.

Built-in Skills

Conduct ships a set of built-in skills bundled with the product. These are not stored in the Knowledge Base and require no configuration:

brand-pack-setup, browser-portal-download, build-dynamic-frontend, create-skill, doc-generator-shared, docx, edit-skill, interactive-html, pdf, pptx, skill-versions, template-discovery, template-library-setup, unique-cli-browser, update-dynamic-frontend-space, user-memory, xlsx

Six of them appear only when the corresponding feature is enabled on the space:

Separately, each connected capability contributes its own unique-cli-* skill automatically — knowledge base search, web search, MCP connectors, sub-agents and others appear when their tool is enabled and disappear when it is turned off. If the agent lists a skill you don't recognize from the set above, this is usually why.

Overriding a Built-in Skill

Any built-in skill can be replaced without a deployment. Create a folder whose name matches the built-in skill name character-for-character, containing a SKILL.md, inside one of the precedence layers — for example company-core/docx/ to replace docx company-wide, or team-credit/pptx/ for one desk only. The Knowledge Base version then fully replaces the bundled skill.

Two caveats: put the override inside a layer folder, not a plain folder — a plain folder overlays files on top of the bundled skill rather than replacing it cleanly. And if the name does not match exactly, you get an additional skill rather than an override.

Company-Specific Skills

In addition to the built-in skills, we recommend organizations develop their own skills for their most common workflows. The agent can help — users can instruct Conduct to create skills directly in a session.

We recommend developing company-specific skills where:

• Intellectual property is involved (proprietary models, analysis frameworks)

• Secure system access is required (internal APIs, databases)

• House style needs enforcement (memo formats, branding, disclaimers)

• Regulatory compliance demands specific output structures

Configuring Skills for a Space

Step 1: Set Up the Skills Folder in the Knowledge Base

1. Decide on a name for the Conduct skills folder. We recommend sticking with the default skills-conduct

2. Navigate to the Knowledge Base in the admin interface and open the Conduct skills folder

3. Create a folder structure inside it following the layer pattern above

4. Upload skill folders (each containing a SKILL.md and supporting files)

Step 2: Enable the Skills Tool on the Space

1. Open the Conduct space in the admin interface

2. Enable the Skills tool

3. Check the Conduct skills folder in the Skills picker. New spaces default to skills-conduct; you can pick a different folder if this space should load a different set, but it is usually preferable to use Every skill in the selected folder and its subfolders is loaded for the space.

4. Optionally toggle skill creation on the same tool — this is what lets users save new skills from a session, and what surfaces create-skill, edit-skill, and skill-versions

Step 3: Verify Skills Are Loading

1. Open the space as a user

2. Ask the agent: "What skills do you have available?"

3. The agent should list the skills loaded from the configured folder

Writing Effective Skills

The Description Matters Most

The description field in SKILL.md front matter is the primary trigger for skill selection. The agent reads it to decide whether to use the skill. Make it specific:

Good: "Extracts key data points, tables, and metadata from financial documents (10-K, DDQ, side letters). Use when the user uploads a financial PDF or asks to parse/extract from a regulatory document."

Bad: "Processes documents"

Instruction Patterns

• Explain the why — Tell the agent why each instruction matters, not just what to do. This helps it make judgment calls on edge cases.

• Include examples — Show input/output pairs so the agent understands the expected format.

• Define output structure — Specify the exact template or format the output should follow.

• Handle errors — Describe what to do when data is missing, malformed, or ambiguous.

• Keep it focused — One skill per workflow. Don't bundle unrelated capabilities.

Bundled Scripts

For deterministic or repetitive tasks, include executable scripts in scripts/. The agent can run these directly instead of writing code from scratch — faster and more reliable.

Testing Skills

We recommend an iterative testing loop:

1. Configure the full set of tools, MCP servers, sub-agents, and skills the agent will need in production

2. Run representative tasks end-to-end in the UI — observe tool selection, skill triggering, and output quality

3. Provide the agent feedback based on where the agent hesitated, misrouted, or produced poor output

4. Instruct the agent to update the existing skill from its own learnings and your feedback

Skill descriptions and instructions are the highest-leverage thing to tune — they determine whether the right skill fires and how well it executes.

FAQs

Q: How many skills can a space have?
A: There's no hard limit, but all skills are loaded at the start of a turn. A very large number increases setup time and can dilute the agent's ability to select the right one. Focus on quality over quantity.

Q: Can different spaces use different skills?
A: Yes — each space's Skills tool points at its own Knowledge Base folder. You can configure different skill sets for different teams, use cases, or client deployments. A single shared folder can also serve several spaces, with space-<space name> folders specializing individual ones.

Q: How do I update a skill?
A: Edit the skill files in the Knowledge Base. Users pick up the new version on a subsequent message — no space reconfiguration needed. Skills are cached briefly for performance, so allow a few minutes before concluding an edit didn't take.

Q: Can users override or disable a skill?
A: Users cannot disable a skill, but they can override one. Saving a skill through create-skill or edit-skill writes it to that user's personal-<userId> folder — the highest precedence layer — so a same-named skill supersedes the company, team, or built-in version for that user only. Editing a shared skill forks a private copy rather than changing it for everyone. Users can also steer the agent explicitly in conversation (e.g. "don't use a template — give me raw output").

Q: What happens if two skills overlap?
A: It depends on whether the names collide. Skills with different names are selected by description: the agent picks whichever description best matches the task, so overlapping descriptions can make selection unpredictable — keep them distinct and specific. Skills with the same name never compete, because precedence resolves them before the agent sees them: only the highest layer's version is loaded.

Q: Can skills access the Knowledge Base or external APIs?
A: A skill is a package of instructions and files, which may include executable scripts. It has no special access of its own, but it can direct the agent to use the tools the space already has — searching the Knowledge Base, calling an MCP server, and so on — as part of its workflow.

Q: Does any of this change when the space is called via the API?
A: No. A Conduct space is reachable through the Unique API like any other space, and API calls inherit the space's configuration — the same skills, tools, and precedence rules apply. The API flow is asynchronous: post a message, then poll for the reply. See the Unique API reference for endpoint details.

Security

• Skills are stored in the Knowledge Base and subject to the same access controls as any KB content

• Skills are downloaded over the platform's internal APIs — no external network access required

• Skill writes from a session always land in the author's own personal-<userId> folder. A user can never modify a shared skill or another user's skill, only shadow it for themselves.

• Other users' personal-* folders are never loaded into a session, independently of Knowledge Base permissions

• Skills cannot escalate the agent's permissions — they can only guide how the agent uses the tools the space already grants it